This record contains the supplementary materials and reproducible analysis assets for the manuscript “Real-time biomass estimation in high-density yeast fermentations using soft-sensor modeling”. We provide paired online–offline biomass datasets and the full statistical workflow used to develop and validate regression-based soft sensors that translate nonlinear OD₈₆₀ (online, in situ) into conventional biomass indicators (DCW and OD₆₀₀) for high-density fed-batch fermentations of three yeast species: Pichia pastoris, Yarrowia lipolytica and Kluyveromyces marxianus. Model performance was assessed using nested leave-one-batch-out cross-validation (LOBO-CV) to avoid information leakage across batch-correlated observations, and model selection prioritized parsimony using a practical equivalence tolerance consistent with typical measurement uncertainty. The deposit includes: (i) raw paired measurements (173 points from 19 independent fed-batch runs), (ii) breakpoint stability summaries across cross-validation folds, (iii) tolerance sensitivity analyses supporting robust model selection, (iv) baseline linear model benchmarking, (v) complete cross-validated performance tables for all candidate models, (vi) residual/heteroscedasticity diagnostics, and (vii) an annotated R script to reproduce all tables and figures reported in the manuscript. Authors: Ana G. Del Hierro; José Luis Checa-Barrera; Juan A. Moreno-Cid; Eoin Casey (corresponding).Affiliations: University College Dublin; BiOrbic (Research Ireland Centre for Bioeconomy); Bionet (Spain).
Hierro et al. (Thu,) studied this question.